Plain-language answer
“Small world!” encounters — meeting a stranger who turns out to share a mutual friend, an old classmate, or a workplace connection — feel rare because we picture searching for a specific person across the entire population of strangers. But real social networks are not a random, even spread of connections: they combine tight local clusters (friend groups, schools, workplaces) with a small number of long-range links that connect distant clusters to each other. That combination is exactly what makes short chains of connection between “random” strangers far more common than intuition suggests.
Why it matters
If everyone’s friends were a random, unrelated sample of the population, a “we have a mutual friend” encounter really would be a strikingly rare coincidence. But real networks are strongly clustered by geography, profession, education, and interest, and laced with occasional links that jump between clusters (a person who moved cities, changed careers, or attended a large conference). Modeling that structure explains why short paths between “random” people are common, without requiring anything more than ordinary social behavior.
Worked example: two ways of studying the small world
In 1967, Stanley Milgram ran an experiment mailing letters to a starting group in the American Midwest, asking each recipient to forward the letter toward a specific target person (a stockbroker in Boston) via someone they knew personally, one step at a time. Chains that reached the target did so in a median of about five to six intermediate steps, giving rise to the popular phrase “six degrees of separation.” The study is historically important as an early empirical probe of network structure, but most started chains never actually reached the target and were excluded from the average — a real limitation of the original method that later researchers have scrutinized closely.(Milgram, 1967)
Three decades later, mathematicians Duncan Watts and Steven Strogatz gave the phenomenon a rigorous mechanism. Starting from a network where everyone is connected only to their near neighbors (highly clustered, but requiring many steps to reach a distant node), they showed that rewiring just a small fraction of connections to distant, random nodes collapses the typical path length dramatically — while barely reducing how clustered the network remains locally. A network can therefore be simultaneously “tightly-knit everywhere” and “only a few steps across,” and it takes surprisingly few long-range shortcuts to produce that effect.(Watts & Strogatz, 1998)
Common misconception
“What are the odds that this specific stranger and I share a friend?” treats the encounter as if it required searching the whole population for one exact match. The small-world network structure means the real question is closer to “given how tightly clustered and how well-connected most social networks are, how many strangers does a typical person share at least one connection with?” — and that number is often much larger than people assume, because clustering plus a few long-range links multiplies the reach of a personal network far beyond its literal size.
Limits and open questions
Watts and Strogatz’s model is a simplified, idealized network, useful for showing that the small-world structure can produce short paths, not for predicting the exact path length between two specific real people, whose actual social graph is shaped by far messier factors (geography, class, migration, platform-specific online networks) than a uniform lattice with random rewiring. Milgram’s original figures are also weaker evidence than they are often given credit for, given the high proportion of incomplete chains in the original data.
Related
- Independence and dependence covers the general point that shared social context breaks independence assumptions — small-world structure is a specific, well-studied version of that idea.
Related interactive
- Social network crossings (in development) will let readers explore a simplified small-world model directly.
Key takeaways
- Real social networks combine tight local clustering with a small number of long-range connections — a structure that produces short paths between “random” strangers far more often than a naive random-network model would.
- Milgram’s original small-world letter experiment is historically important but methodologically limited; Watts and Strogatz later gave the phenomenon a rigorous mathematical mechanism.
- A “mutual friend” encounter with a stranger is not evidence of a meaningful connection — it is close to the expected behavior of ordinary, clustered-plus-shortcut social networks.